Lifecycle
Morse AI’s screens change shape as your account matures. A brand-new account sees activation prompts; an account with a little data sees “forming” states; a steady account sees the full evidence-backed operating layer. This isn’t cosmetic — it’s the product refusing to show verdicts it can’t yet stand behind.
Maturity is driven by volume, not by the calendar
The intuition “older account = more features” is close but wrong in an important way. Morse AI promotes a capability when it has enough data, not when enough time has passed. A high-volume account can reach the full operating layer within its first hour; a low-volume one can stay in “forming” for weeks.
Each of the four capabilities — cost, reliability, quality, context — has its own sufficiency
level: none, forming, or verdict-ready. A capability only shows verdicts once it’s
verdict-ready; below that, it shows a gate line explaining what it’s still waiting for.
Those levels roll up into an overall account stage:
| Stage (code) | Doctrine name | What it means |
|---|---|---|
none | S0 — Empty | Nothing instrumented. The screen’s job is activation: get your first trace. |
forming | S1 / S2 — First data / baseline forming | Some capabilities have data; each gates independently. |
ready | S3 — Operating | At least three of four capabilities are verdict-ready; the full operating layer renders. |
The S0–S3 names come from Morse AI’s internal product doctrine; the code speaks in none /
forming / ready plus the four per-capability levels. Both describe the same behavior.
What “enough data” means, concretely
The thresholds are explicit. A few examples:
- Cost becomes verdict-ready at 1,000 priced runs, or 200 priced runs across 5+ days of spans; it starts forming at 25 priced runs.
- Reliability is verdict-ready at 200 runs over 5+ days; forming at 50.
- Quality is verdict-ready when 80% of runs carry outcomes and at least 50 are scored; forming at 40% coverage.
- Context is verdict-ready at 200 profiled LLM spans; forming at 25.
Because each capability gates on its own counter, it’s normal to see, say, Reliability showing verdicts while Cost still says “forming — 25 to go.” Every gate line names its own threshold, so you always know what unlocks it.
Demo mode renders a full account
The demo workspace shows the steady operating layer — but it gets no special-casing in the lifecycle logic. Its seeded dataset is simply large enough to cross the same thresholds as any real account. That’s the point: demo mode is a truthful preview of what your own account looks like once it has data, not a mock-up.
Related
- Honest numbers — the gate lines this page’s stages produce.
- Dashboard — where the Org Brief and capability cards adapt to your stage.